Enhancing Book Recommendation Systems: A Deep Dive into Weighted Alternating Least Square (WALS) and Neural Collaborative Filtering (NCF) with Feature Optimization
K. Kavitha, Sankar Murugesan · International Journal of Electronics and Communication Engineering · 2024
A Recommendation System (RS) is a kind of data filtering framework that can forecast user preferences or ratings for various categories, including music, movies, books, social media tags, books, and general products. A book recommendation system is essential to connect readers with appropriate books, encourage a love of reading, and preserve an exciting literary community. With the rise of online bookstores and digital libraries, readers would not be able to discover their next outstanding literary adventure without personalized book recommendations. This work primarily aims to present a comparative analysis of the performance of suggested book recommendation systems employing the Neural Collaborative Filtering (NCF) approach with feature optimization and the Weighted Alternating Least Square (WALS) approach. The proposed models were evaluated on the GoodBooks-10Kdataset. Root Mean Square Error (RMSE) values were employed to compare the models’ performances. A system that is better at forecasting user behavior will provide a more satisfying and customized reading experience; a decreased RMSE score indicates this. The simulation outcome indicates that the suggested method produced excellent outcomes with significantly lower RMSE values. It also demonstrates that NCF with feature optimization exhibits superior recommendation performance regarding RMSE, outperforming WALS consistently with lower values. This outcome demonstrates how the recommended techniques can enhance the effectiveness of book recommendations and help users select books that are more compatible with their own tastes.